Using thermocouples with T7

Hi Nick,

Ground loops are big problem when working with thermocouples. The best way to deal with this is to switch to differential readings. The differential reading is not prone to most grounding issues, and usually results in lower noise due to common mode rejection.

Check out our Thermocouple app-note for common pitfalls and solutions: https://labjack.com/support/app-notes/thermocouples

I couldn't agree more.  Differential input is crucial both to avoid ground loops and to provide good common-mode rejection on tiny signals.  However, the T7, like many well designed DAQs, has high input impedance, so static charges can build up if the TC beads aren't in contact with a ground.  The result is an irritating tendency for the signal to float up to the rails of the high gain input, causing all manner of unexpected bizarre signal behavior due to partial (only one of the inputs) or full saturation (resulting in an unexpected intermittent zero signal).

In addition to using differential input, it has worked well for me in the past to put a >100k resistor between the negative lead and ground, so there is sufficient resistance to render ground loops irrelevant, but static charges have a place to discharge.  I currently have an array of type K TCs on my T7, and they give excellent measurements.

Great added comments CMartin.  Some customers also add a 100k or 1M resistor from the positive input to VS.  That way a floating input (no thermocouple connected) will give a railed invalid reading that software can easily notice.

I too utilize various thermocouples on a T7-pro. The calculated temperature is excellent, as confirmed by a NIST traceable calibrator that we use here. Some other thoughts:

Use as high of a resolution index and settling time as you can afford. Also set the gain to an appropriate level. A certain amount of smoothing/averaging will also help significantly (try an exponential filter in a lua script and then interogate the filtered result https://en.wikipedia.org/wiki/Exponential_smoothing)

Place a small capacitor (~0.1 uF) across the terminals.

Three wire thermocouples can be made where the center wire is ground (or at some Vref), the outer two go to a differential input pair.

Instead of directly touching the surface electrically isolate the bead with a thermal pad (http://www.digikey.com/products/en?keywords=ber130). Though if you're at low enough temp that you can get away with those you might want to look at something like the LM34 instead of thermocouples...

Also one question regarding the 1 MOhm to Vs rail: are carbon comp resistors sufficient or do people have better luck with lower noise types? It seems that as long as the thermocouple has a low resistance it shouldn't matter...

Regards,

- Jason

Good question.  We will run some tests.

I used the analog input tab in Kipling to set AIN0 to type K and then watched the raw voltage value displayed just to the right of "AIN0".  I used defaults (Range=0.1, ResIndex=Auto=9 for my T7-Pro) but set NegativeCh=AIN1.

With AIN0 & AIN1 both jumpered to GND I saw almost no noise.  Maybe about 1uV.

With a type K thermocouple connected to AIN0 & AIN1, with a short jumper from AIN1 to GND, I saw similar noise.  Maybe more like +/-2uV, but that could be actual temperature fluctuations of air currents in my office.

With a type K thermocouple connected to AIN0 & AIN1, with a 1M resistor from AIN1 to GND, I saw similar noise.  The large resistor does not seem to create any noticeable added noise in this scenario.